TranscriptCleanup.pl

From radmind

(Difference between revisions)
Jump to: navigation, search
(Initial Content.)
Current revision (02:57, 4 September 2008) (edit) (undo)
(Added a --dangling option that checks for files without transcripts)
 
Line 17: Line 17:
my $command_dir = "/var/radmind/command";
my $command_dir = "/var/radmind/command";
my $config_file = "/var/radmind/config";
my $config_file = "/var/radmind/config";
 +
my @file_dirs = ( "/var/radmind/file", "/var/radmind/file/relative");
# Sanity check the first parameter, and make sure it is given
# Sanity check the first parameter, and make sure it is given
my $opt = shift;
my $opt = shift;
-
if (not defined $opt or $opt !~ m/--all|--zero/)
+
if (not defined $opt or $opt !~ m/--all|--zero|--dangling/)
{
{
-
   print "Must use option \"--all\" or \"--zero\"\n";
+
   print "Must use option \"--all\" or \"--zero\" or \"--dangling\"\n";
   exit 1;
   exit 1;
}
}
Line 37: Line 38:
my %K = ();
my %K = ();
my @deep_K = ();
my @deep_K = ();
 +
my @F = ();
Line 44: Line 46:
File::Find::find({wanted => \&wanted_K}, $command_dir);
File::Find::find({wanted => \&wanted_K}, $command_dir);
 +
# Find all the top level .T file dirs
 +
foreach my $file_dir (@file_dirs)
 +
{
 +
  my @globvalue = glob ("$file_dir/*.T");
 +
  foreach my $gv (@globvalue)
 +
  {
 +
      push (@F, $gv) if ( -d "$gv" );
 +
  }
 +
}
# Parse the radmind config file to find all the starting K files
# Parse the radmind config file to find all the starting K files
Line 67: Line 78:
# Now, simply print out all the references for all the transcripts
# Now, simply print out all the references for all the transcripts
-
foreach my $f (keys %K)
+
if ($opt =~ m/--zero|--all/)
{
{
-
   my $i = $K{$f}{'ref'};
+
   foreach my $f (keys %K)
-
  print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
+
  {
-
   print("$i = $f\n") if ($opt =~ m/--all/);
+
      my $i = $K{$f}{'ref'};
 +
      print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
 +
      print("$i = $f\n") if ($opt =~ m/--all/);
 +
   }
 +
  foreach my $f (keys %T)
 +
  {
 +
      my $i = $T{$f}{'ref'};
 +
      print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
 +
      print("$i = $f\n") if ($opt =~ m/--all/);
 +
  }
}
}
-
foreach my $f (keys %T)
+
elsif ($opt =~ m/--dangling/)
{
{
-
   my $i = $T{$f}{'ref'};
+
   foreach my $f (@F)
-
   print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
+
   {
-
  print("$i = $f\n") if ($opt =~ m/--all/);
+
      my $cpath = $f;
 +
      $cpath =~ s/file/transcript/;
 +
      #next if ( defined $T{$cpath} and $T{$cpath}{'ref'} > 0 );
 +
      print "/bin/rm -rf $f;\n" unless ( defined $T{$cpath} );
 +
     
 +
  }
}
}
Line 92: Line 117:
     }
     }
}
}
-
 
Line 103: Line 127:
     }
     }
}
}
 +
#-------------------------------------------------------------------------------
#-------------------------------------------------------------------------------

Current revision

#!/usr/bin/perl -w
# Description : This finds all the transcripts, command files, and reads the
#               config file and then determines how many times each transcript
#               is referenced.  Those transcripts that are referenced zero times
#               should be safe to remove.  This script attempts to deal
#               correctly with "K-in-K" files.
#
#     License : GNU GPL version 3
#      Author : Paul Ortman <portman@goshen.edu>
#        Date : Feb 2008
################################################################################
use strict;
use File::Find ();

my $trans_dir = "/var/radmind/transcript";
my $command_dir = "/var/radmind/command";
my $config_file = "/var/radmind/config";
my @file_dirs = ( "/var/radmind/file", "/var/radmind/file/relative");

# Sanity check the first parameter, and make sure it is given
my $opt = shift;
if (not defined $opt or $opt !~ m/--all|--zero|--dangling/)
{
   print "Must use option \"--all\" or \"--zero\" or \"--dangling\"\n";
   exit 1;
}

# for the convenience of &wanted calls, including -eval statements:
use vars qw/*name *dir *prune/;
*name   = *File::Find::name;
*dir    = *File::Find::dir;
*prune  = *File::Find::prune;


# Hashes for the Transcripts, Kommand files, and a "K-in-K" queue
my %T = ();
my %K = ();
my @deep_K = ();
my @F = ();


# Traverse desired filesystems to find _all_ the transcripts and command files
# in the filesystem
File::Find::find({wanted => \&wanted_T}, $trans_dir);
File::Find::find({wanted => \&wanted_K}, $command_dir);

# Find all the top level .T file dirs
foreach my $file_dir (@file_dirs)
{
   my @globvalue = glob ("$file_dir/*.T");
   foreach my $gv (@globvalue)
   {
      push (@F, $gv) if ( -d "$gv" );
   }
}

# Parse the radmind config file to find all the starting K files
parse_config_file();


# Parse each used command file from the config, and keep track of 'K-in-K' in an
# additional array.
foreach my $f (keys %K)
{
   parse_command_file($f) if ($K{$f}{'ref'} > 0);
}


# Work through every element in the array until the array is empty - fake
# recursion.
while (my $f = pop(@deep_K))
{
   #print "$f\n";
   parse_command_file($f) if ($K{$f}{'ref'} > 0);
}


# Now, simply print out all the references for all the transcripts
if ($opt =~ m/--zero|--all/)
{
   foreach my $f (keys %K)
   {
      my $i = $K{$f}{'ref'};
      print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
      print("$i = $f\n") if ($opt =~ m/--all/);
   }
   foreach my $f (keys %T)
   {
      my $i = $T{$f}{'ref'};
      print("$i = $f\n") if ($opt =~ m/--zero/ and $i == 0);
      print("$i = $f\n") if ($opt =~ m/--all/);
   }
}
elsif ($opt =~ m/--dangling/)
{
   foreach my $f (@F)
   {
      my $cpath = $f;
      $cpath =~ s/file/transcript/;
      #next if ( defined $T{$cpath} and $T{$cpath}{'ref'} > 0 );
      print "/bin/rm -rf $f;\n" unless ( defined $T{$cpath} );
      
   }
}

exit;



#-------------------------------------------------------------------------------
sub wanted_T 
{
    if (/^.*\.T\z/s && -f "$name")
    {
       $T{"$name"}{"ref"} = 0;
    }
}


#-------------------------------------------------------------------------------
sub wanted_K 
{
    if (/^.*\.K\z/s && -f "$name")
    {
       $K{"$name"}{"ref"} = 0;
    }
}


#-------------------------------------------------------------------------------
sub parse_config_file
{
   open(my $CONFIG, "< $config_file") or die $!;

   while(<$CONFIG>)
   {
      chomp(); # cut off the newline

      next if (m/^\s*#/); # skip blank lines or commented lines
      next if (m/^\s*$/); # 
   
      my ($comp, $file) = split(/\s+/); # tokenize the line

      next unless ($file =~ m/\w/); # make sure something is listed as the file

      # Count the references
      $K{"$command_dir/$file"}{"ref"}++ if ($file =~ m/\.K$/);
      $T{"$trans_dir/$file"}{"ref"}++ if ($file =~ m/\.T$/);
   }
   close ($CONFIG);
}

#-------------------------------------------------------------------------------
sub parse_command_file
{
   my $target = shift;
   open(my $CONFIG, "< $target") or die $!;

   while(<$CONFIG>)
   {
      chomp(); # cut off the newline

      next if (m/^\s*#/); # skip blank lines or commented lines
      next if (m/^\s*$/); # 
   
      my ($type, $file) = split(/\s+/); # tokenize the line

      next unless ($file =~ m/\w/); # make sure something is listed as the file

      # Count the references
      if ($type =~ m/k/ and $file =~ m/\.K$/)
      {
         $K{"$command_dir/$file"}{"ref"}++;
         push @deep_K, "$command_dir/$file";
      }

      if ($type =~ m/p|n/ and $file =~ m/\.T$/)
      {
         $T{"$trans_dir/$file"}{"ref"}++;
      }
   }
   close ($CONFIG);
}
Personal tools